A Study on Spatiotemporal Downscaling Methods for Chlorophyll-a Concentration in Taihu Lake Based on Remote Sensing Data from Sentinel-2 MSI and COMS-1 GOCI

Taihu Lake is a large lake with high levels of eutrophication. Cyanobacterial outbreaks significantly affect the ecological environment and socioeconomic development. The chlorophyll-a (Chl-a) concentration, which is crucial for monitoring eutrophication, can be obtained through remote sensing inversion, and the random, sudden, and complex changes impose stringent requirements on the monitoring scale. However, single remote sensing images often fail to meet both the high temporal and spatial resolution requirements for Chl-a monitoring. This study took Taihu Lake as the research object, combined COMS-1 GOCI (1 h/500 m resolution) and Sentinel-2 MSI (5 d/10 m resolution) inverted Chl-a data, and developed a precorrection-based spatiotemporal downscaling method (PC-STDM). After eliminating systematic bias, the model used temporal weighting downscaling (TWD) and regression trend assessment downscaling (TRAD) methods to downscale the inverted Chl-a data, improving the temporal resolution of the Sentinel-2 MSI Chl-a inversion data from 5 d to 1 h. The verification resulted in an average R2 of 0.87 between the COMS-1 GOCI and Sentinel-2 MSI Chl-a data after adaptive correction. A comparison with the measured Chl-a data yielded a maximum fitting coefficient of 0.98, verifying the credibility of the model. The downscaled Chl-a concentration data detailed hourly changes and development trends, providing support for water quality monitoring in the Taihu Lake area.